Python Logging

Also Known As

TL;DR

Explanation

The logging hierarchy (DEBUG, INFO, WARNING, ERROR, CRITICAL) lets you filter verbosity by environment. Loggers are named (logging.getLogger(__name__)) creating a hierarchy matching the module structure. Handlers direct output to files, streams, or external services. Formatters define the output format. Never use print() for operational logging — it has no severity, no filtering, no structured output, and cannot be silenced in production without code changes.

Common Misconception

Why It Matters

Common Mistakes

• Using the root logger (logging.info()) instead of a named logger — root logger pollution breaks third-party library logging.
• Configuring logging in library code — libraries should add NullHandler only; applications configure logging.
• f-string interpolation in log messages: logger.debug(f'Value: {x}') — evaluated even when DEBUG is disabled; use % formatting.
• Not setting log level for third-party libraries — they default to WARNING; chatty libraries need explicit suppression.

Code Examples

# print() for logging — no severity, no filtering:
def process_order(order_id):
    print(f'Processing order {order_id}')  # No level, no off switch
    result = do_processing()
    print(f'Result: {result}')  # Pollutes production output

# Named logger with structured output:
import logging
logger = logging.getLogger(__name__)  # Named after module

def process_order(order_id: int) -> None:
    logger.info('Processing order', extra={'order_id': order_id})
    try:
        result = do_processing()
        logger.debug('Processing result', extra={'order_id': order_id, 'result': result})
    except ProcessingError:
        logger.error('Order processing failed', extra={'order_id': order_id}, exc_info=True)

References

• ↗ https://docs.python.org/3/howto/logging.html

Tags

Related Terms